Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Journal of Hydrology, 37 (1978) 309--322 309

© Elsevier Scientific Publishing Company, Amsterdam -- Printed in The Netherlands

[4]

MODEL FOR SIMULATING SOIL-WATER CONTENT CONSIDERING


EVAPOTRANSPIRATION

ABAS AFSHAR and MIGUEL A. MARINO


Department of Land, Air and Water Resources, and Department of Civil Engineering,
University of CaL!fornia, Davis, CA 95616 (U.S.A.)
(Received May 25, 1977; revised and accepted September 26, 1977)

ABSTRACT

Afshar, A. and Marino, M.A., 1978. Model for simulating soil-water content considering
evapotranspiration. J. Hydrol., 37: 309--322.

A mathematical model is developed to predict the soil-water content profile under


transient field conditions. It takes into account a variable transpiration rate and a nonuni-
form initial soil-water content. Rainfall, irrigation, and evaporation are treated as sources
of nonuniform potential surface flux. The possibility of no flux at the soil surface is also
included in the model. Solutions to the model are obtained numerically by a finite-dif-
ference scheme involving predictor--corrector equations. The scheme results in a tridiag-
onal set of simultaneous equations which can be solved rapidly in a digital computer.
A step-by-step description of the computational procedure is included. Application of
the model to some field conditions and comparison of the results with field-measured
data shows very good agreement. Sensitivity of the model to changes in rooting zone, root
density function, and depth to the bottom boundary is also analyzed.

INTRODUCTION

The availability of water for plant roots is an important topic which has
been explored by a number of investigators (e.g., Gardner, 1960, 1964;
Whisler et al., 1968; Molz and Remson, 1970; Nimah and Hanks, 1973a;
Feddes et al., 1974). Recently, attention has been given to irrigation manage-
ment for control of quality of irrigation return flow, salt concentration in the
root zone due to irrigation water, and uptake by plant roots from a free
saline water table (King and Hanks, 1973). Such management strongly de-
pends upon knowledge of water and salt movement through the root zone of
the crops. Prediction of available moisture for plant roots also has a signifi-
cant effect in irrigation management.
Soil-water flow to plant roots has been studied by a number of investiga-
tors. The research has followed basically two different approaches. Philip
(1957), Gardner (1960) and Molz et al. (1968) considered the radial flow of
water to a single root. The term "microscopic approach" is used to designate
this approach. In this case, a single root is assumed to be represented by a
310

narrow and infinitely long cylinder of constant radius which absorbs water.
Because any experiment based on this approach deals with a single root, it i s
not directly applicable to the field. In contrast with the microscopic approach,
other investigators (Gardner, 1964; Whisler et al., 1968; Molz and Remson,
1970; Nimah and Hanks, 1973a; Feddes et al., 1974) have focused their at-
tention on the removal of moisture from a differential volume of soil as a
whole, without considering the effect of individual roots. The term "macro-
scopic approach" is used to designate this approach.
Gardner (1964) employed a mathematical model to describe water uptake
by roots. In laboratory experiments representing steady-state flow conditions,
a measure of root concentration at each depth increment was calculated. The
basic idea of Gardner's work was to actually find a set of numbers for root
concentration associated with each depth increment. Whisler et al. (1968)
applied a numerical technique to the steady-state flow equation for evapo-
transpiration from a vertical soil column. No attempt was made to check the
result to the solution of the model vs. experimental data. The objective of
their paper was to demonstrate that solutions to the steady-state flow equa-
tion with a (negative) source term can be obtained by numerical means.
Molz and Remson (1970) developed a mathematical model to describe water
movement to plant roots. Because their model did not consider a flow con-
dition at the soil surface, the effects of irrigation, rainfall, and evaporation
were not taken into account. In addition, Molz and Remson (1970) assumed
a constant average transpiration rate and a uniform initial moisture content.
The model was tested against experimental laboratory data from Gardner and
Ehlig (1962) and Gardner (1964). The model's prediction goes at most for
eight days, and y e t even on the eighth day, there is poor agreement. Nimah
and Hanks (1973a, b) developed a one-dimensional model to predict soil-
water content and tested it with field data. For the available data, the maxi-
mum relative deviation did not exceed 23%. A relatively poor comparison
was found between predicted and measured moisture content profiles at 24 h
after irrigation which the authors attributed to hysteresis. The possibility of
a zero flux at the soil surface was not included in the model. Feddes et al.
(1974) slightly modified the soil-moisture model of Nimah and Hanks (1973a)
and tested it with a new set of data. The modification consisted of a combi-
nation-type of equation which was used to calculate maximum possible flux
due to transpiration and the replacement of the root distribution function of
Nimah and Hanks (1973a) by another measure experimentally found by
Feddes (1971). The authors remarked that this measure has a linear relation
with root mass and that the proportionality constant may be used to cali-
brate the model. Apparently, from a literature survey, Feddes et al. (1974)
found that relative root mass is an exponentially decreasing function with
depth. However, data presented by Nimah and Hanks (1973a, b) were ex-
cluded from the survey. The authors also remarked that finding this kind of
measure for root distribution is expensive as well as experimentally difficult
to perform. Feddes et al. (1976) attempted to describe the sink term as a
311

function of soil-water content. As we will show later in this paper, the func-
tion of Feddes et al. (1976) underestimates the soft-water content profile and
in many cases does not yield results that are comparable to those measured
in the field.
The objectives of this paper are: (1) to investigate the performance of the
sink term of Molz and Remson (1970) b u t under field conditions by con-
sidering rainfall, irrigation and evaporation as sources of surface flux and
using the root density function of Nimah and Hanks (1973b); and (2) to
study the effect of evaporation on the soil-water content profile by account-
ing for and disregarding evaporation in two different boundary-value prob-
lems. The first objective will be achieved by comparing the soft-water content
profiles predicted by the solution to the proposed model with those pre-
dicted by the solution to other types of sink term models as well as with
field-measured data. Sensitivity of the proposed model to various parameters
will also be analyzed. It should be noted that the proposed model does not
account for plant control on water uptake when soil water is a limiting factor.

MODEL DESCRIPTION AND METHOD OF SOLUTION

In the macroscopic approach, water-removing roots may be represented by


an extraction term in the general flow equation. If the rate at which water is
removed from a differential volume of soil is denoted by W, the one-dimen-
sional flow equation can be written as:
30 /3t = (3/bz) (K (3¢/3z)-- W (1)
in which 0 = moisture content; K = hydraulic conductivity; ¢ = hydraulic head;
W = rate of extraction; z = vertical coordinate, directed downward; and t =
time coordinate.
Realizing that the extraction rate from the r o o t zone is equal to the trans-
piration rate, Molz and Remson (1970) considered transpiration to be distrib-
uted according to diffusivity and effective root density. Effective r o o t den-
sity is related to the fraction of roots that is effective in absorbing water from
soft. The extraction term of Molz and Remson is in the form:
v
W(z, O) = TR(z) D(O)/ f R(z) D(O) dz (2)
o
where R (z) = effective root density; T = transpiration rate; and V = root
depth. It should be noted that W(z, 0 ) is defined so that the total extraction
rate in the root zone equals the transpiration rate, that is:
V
T= f W(z,O)dz
o

For computational purposes, r o o t density (weight of roots per bulk volume


of soft) is taken as a measure of effective root density (Nimah and Hanks,
1973b).
312

Combining eqs. 1 and 2 and using diffusivity yields:

aO(z,t) aD(O) aO(z,t) a20(z,t) ag(o) TR(z)D(O)


- +D(O) (4)
at az az az: Oz /rVR(z)D(O )dz
J
0

in which D = K (dh/dO) = diffusivity; and h = pressure head (negative if soil


is unsaturated).
The moisture content model under consideration is given by eq.4 subject
to initial and boundary conditions:
0 (z, t) = 0 (z, O)input O<~z<~L t=o (5)
D [~0 (z, t)/az] - - K ( O ) = 0 z =0 tk < t < tt (6)
0 (0, t) = 0comp Z= 0 t m < t < tn (7)
0 (z, t) = 0 (z)sat z =L t t> 0 (8)
where 0 (z, 0)input represents known values of 0 (z,0); 0 (Z)sat is the saturated
soil-moisture content for a given soil; tk and t I represent time intervals in which
the flux at the soil surface is zero; and tm and tn represent time intervals in
which some potential flux (e.g. potential evaporation or effective precipitation)
is specified at the soil surface. It should be noted that tk /> 0 and tm >~ 0
whereas tl > 0 and tn :> O. 0 comp represents the c o m p u t e d value of the soil-
moisture content at the soil surface. The technique employed to calculate
0 comp will be discussed later.
To investigate the effect of evaporation on the soil-moisture profile, we are
interested in the solution of two boundary-value problems. One of these
problems is given by eq. 4 subject to eqs. 5, 7 and 8. In this case, evapora-
tion as well as irrigation and rainfall are considered to be the possible sources
of potential flux at the soil surface. The remaining boundary-value problem
is described by eqs. 4--8. In this latter case, evaporation is disregarded.
The soil-moisture flow problems under consideration are solved numeri-
cally by a finite-difference scheme involving the predictor--corrector technique
of Douglas and Jones (1963). The predictor and corrector equations, respec-
tively, for eq. 4 are:
Oi+ld+~/~ --2 0 i,y+v2 + Oi-y+v2 1 [0 i,y+i/~--~i,j~
- (~z) 2 - Di, ] ~ -A~ I/

--.D !" (Oi+1,j--Oi--l,jl +K~


t,j + TjRi 1
Di, j \ 2Az / Di,j f V R i D i , j dz
0
(9)
and

! [ 0 i + 1 , j + 1 -- 20i,j+ 1 + Oi--l,j+ 1 + 0i+1, j -- 20i, j + Oi--l,j. 1


I
2L (Az)2 (~z) ~ 2
313

_ 1 Oi,j+l--Oi, j Di,j+V20i+ld+V 2 +K~j+~/2 TjR i


Di,j+V~
( At ) Di,j+l/2 ( °i
2Az
lJ+ 1 +
'/ Di,j+V2 / V
R iD4jdz
(I0)
where D' = 8D(O )l~z and K ' = ~K(O )laz. Since D is not a monotonically in-
creasing function of 0 (see Fig.5), using 8D/Sz instead of 8D/80 would help
us choose the proper depth increment Az and thus avoid unreliable results.
The general computational procedure involves several steps. A step-by-step
description of the program is as follows:
Step 1. The program reads and prints the input data and boundary con-
ditions. These data include tables of conductivity and soil-water pressure
head as a function of water content, and root density as a function of depth.
Also included are values of potential evaporation and transpiration, irriga-
tion and rainfall, upper and lower limits on pressure head and water content.
The condition at the lower boundary is also specified.
Step 2. Diffusivity, as a function of water content for a given soil, is com-
puted and printed. In this paper, diffusivity is calculated from the relation
D(O) = K(O)Sh(O)/~O using tabular values of 0, h and K reported by King and
Hanks (1973). The small increments in 0 given by King and Hanks (1973),
A0 = 0.01, allowed us to calculate 8h(O)/OO in a finite difference fashion.
Step 3. For any given water content, corresponding values for hydraulic
conductivity and diffusivity are picked up from input-data tables. For values
of moisture content that lie between two successive values given in the input
data, linear interpolation is used to evaluate hydraulic conductivity and
diffusivity. (The use of exponential interpolation adds no difficulty at this
step.) Then by calling the corresponding subroutine, ~K(O )/~z and ~D(O )/~z
are calculated.
Step 4. When flux at the surface is considered, a surface water content is
c o m p u t e d to give the estimated flux at the surface in conformance with
boundary conditions applying at that time. To estimate the flux, the follow-
ing equation is used:
Oo,j+l/2 - 01,j
flux = D1/2,j+ l/2 _ A z + K1A,j+I/2

where 0 0 4+ lh and 01 ,j = water contents at the end and beginning of any half-
time interval at the surface and at depth Az, respectively; and Kv2,j+I/2 =
hydraulic conductivity which is assumed to be constant over the half-time
interval applying between depth 0 and depth Az. The surface water content
is allowed to vary only between saturation and air-dry limits.
It should be noted that the c o m p u t e d flux may be different than the
potential flux. To find the correct flux, a value of 0 0,j+v2 is assumed, be-
cause it has n o t been calculated yet. If evaporation takes place, the assumed
value of 0 0j+ 1/2 is air-dry water content. In the event of precipitation, the
assumed value of 0 o,1+1/2 is saturated water content. By means of iteration,
0 0,j+ 1~ is modified until the c o m p u t e d flux is in conformance with the poten-
tial flux. This c o m p u t e d moisture content, 0 o,j+l/2, is the 0 comp of eq.7.
314

When soil-moisture conditions are such that potential flux cannot occur,
the c o m p u t e d flux will be smaller than the potential flux. For example, im-
mediately before irrigation or rainfall the soil-water content may be so low
that the rate of infiltration equals that of precipitation. If the rate of precip-
itation is sufficiently high, the soil-water content at the soil surface increases
until it reaches saturation. Hereafter, the infiltration rate may be smaller than
the precipitation rate, and hence ponding or runoff occurs. The reverse phe-
nomenon occurs when evaporation takes place.
Step 5. Simpson's one-third rule of integration is used to evaluate the term

f R ( z ) D (0) dz in eq.4. For each time step, a different value of transpiration


0
is used as specified in the boundary condition.
Step 6. The predictor equation, eq.9, is solved by applying the tridiagonal
algorithm, and by repeating step 5.
Step 7. The corrector equation, eq. 10, is solved by using the tridiagonal
algorithm, moisture content is then printed.
Step 8. When total computational time has elapsed, the program reads
values of the potential evaporation and computations are resumed at step 4.
This is to account for evaporation as a source of flux at the surface.
The data needed for the solution of the model are:
(1) Soil properties (i.e., pressure head--water content, hydraulic conduc-
t i v i t y - w a t e r content data) covering the range of water content to be en-
countered in the problem. Values of 0 saturation and ~ air-dry must also be
specified.
(2) Plant properties, root distribution function.
(3) Water content--depth data at the beginning of the computations (initial
condition).
(4) Potential transpiration, potential evaporation, irrigation and rainfall
rates as functions of time for the period of interest (boundary condition).
(5) Presence or absence of a water table at the b o t t o m of the soil profile
(boundary condition).

RESULTS AND DISCUSSION

The consideration of evaporation as a source of flux at the soil surface


gives rise to the solution of eq. 4 subject to eqs. 5, 7 and 8. Moisture content
comparison is made by plotting soil-water content vs. depth in Fig.1. The
solid lines represent the soil-water c o n t e n t profiles predicted by the present
model, the broken lines represent the profiles predicted by the sink term model
of Feddes et al. (1976), the d o t t e d / b r o k e n lines represent the profiles pre-
dicted by the model of Nimah and Hanks (1973b), and the dots are data
which resulted from a field experiment (King and Hanks, 1973). It is evident
that there is good agreement between our predicted soil-water content pro-
files and those measured in the field. For the data reported by King and
Hanks (1973) and for a depth of 30 cm, the maximum percent deviation of
315

SOIL-WATER CONTENT (cm3cm -3)

0 0 0~1 0,2 0.~ 0.4 0.5 OJ 0.2 0.5 0.4 0.5 0 O~l 0.2 0.3 0.4 0.5

I/I ,
~
,o
80
,io /f

180 L
predicted (present model)
predicted (Feddes et al., 1976)
predicted (Nirnah end Honks, 1973b)
oooo measured (King and Honks, 1973)
Fig.1. Comparison between measured and predicted soil-water content profiles at the
indicated simulation times when evaporation is considered for: (a) 48 h after irrigation;
(b) end of irrigation interval; and (c) 48 h after irrigation.

the present model is a b o u t 4.4% and occurs on August 18 while that of the
model of Nimah and Hanks (1973b) is a b o u t 21.8% and also occurs on
August 18. Table I shows the maximum percent deviations for depths of 30,
45, 75 and 135 cm.
The sink term model of Feddes et at. (1976) underestimates the soil-water
content profiles. For instance, in Fig.lb at a depth of 30 cm, the soil-water
content predicted by the model of Feddes et al. (1976) is a b o u t 0.15 where-
as the measured soft-water content is a b o u t 0.245 and that predicted by the

TABLE I

Maximum percent deviation of soil-water content prediction models for the data reported
by King and Hanks (1973)

Depth Present model Nimah and Hanks model


(cm)
max. percent time of max. percent time of
deviation occurrence deviation occurrence

30 4.4 Aug. 18 21.8 Aug. 18


45 14.0 Aug. 19 11.0 Aug. 18
75 21.0 Aug. 18 23.7 Aug. 18
135 13.3 June9 18.8 Aug. 19
316

present model is 0.23. This low soft-water content profile predicted by the
model of Feddes et al. (1976) causes the wetting front formed after irrigation
to advance slowly and 48 h after irrigation results in the soil-water content
profile shown by the broken line in Fig.lc which is in poor agreement with
the field-measured data.
Excluding evaporation from the analysis results in the solution of eq. 4
subject to conditions (5)--(8). Some of the numerical results obtained are
presented in Fig.2. Since potential evaporation for alfalfa was assumed to be

SOIL-WATER CONTENT (cm3cm -3)


o o.i 0.2 0.3 0.4 o . 5 o oJ 0.2 0.3 0.4 0.5
o
20
4o

E 60

80
I00
120
140
160 370 hr "~ 650 hr "k,o
180
predicted (present model)
oooo measured (King and Hanks, 1973)
Fig.2. Comparison between measured and predicted soil-water content proviles at the
indicated simulation times when evaporation is disregarded for: (a) 48 h after irrigation;
and (b) 48 h after irrigation.

10% of evapotranspiration (Nimah and Hanks, 1973b), disregarding evapora-


tion had only a little effect on the soft-water content profiles. For the avail-
able data, however, the agreement between measured and predicted soil-
water content is much better when evaporation is taken into account. It
should be remarked that in situations such as corn starting with a bare soil,
evaporation will be higher than 10% of evapotranspiration and thus cannot
be ignored in calculating soft-water content.
A comparison between predicted and field measured soil-water content
for depths of 30, 45, and 105 cm and for t = 900 h is presented in Fig.3.
Since the soft-water c o n t e n t was initialized after each crop, the plot of soil-
water c o n t e n t vs. time was not extended for the whole period of 2800 h.
As can be visualized in Fig.3, the soil-water content with respect to time
agrees very well with the measured values.
A plot of cumulative upward flow through the b o t t o m boundary (free
317

predicted ( presenl model)

o ooo me0sured (Nimoh 0nd Hanks,lg73b)


O4
~. 0.3 ~
, o2

-o- oo 0 I00 200 300 400 500 ' --70o


600 aoo'

m
~ 04

OI
i~J °2
O0 L.~
0
J
tO0
I
200
I
300
I
400
=
500
I
600
~
700
J
800

0 4 1 (C)
0.2
0.1
0.0 I l I I I I I I
o mo 2oo 300 400 500 600 700 8oo 9oo
TIME ~'hrs)

Fig,3. Comparison between measured and predicted soil-water contents at depths of:
(a) 30 cm; (b) 45 cm, and (c) 105 cm.

water table) as compared to the measured flow between August 5 and Sep-
tember 10 is presented in Fig.4. While evapotranspiration is taking place
some water is moving from the water table to the root zone. After the first
irrigation which occurs from t = 3 1 2 - - 3 2 8 h, a reverse flow takes place.
After the second irrigation (t = 6 1 2 - - 6 2 7 h) which is followed by some rain-
fall, water flows toward the water table until t = 850 h. This reverse flow
is due to the large amount of irrigation and rainfall applied in this period.
The general trend for predicted upward flow is the same as that for measured
flow. While the difference between predicted and measured upward flow
could be due to inaccuracy of the model and/or data, the agreement is prob-
ably sufficiently good for many practical purposes.

u predicled (present model)


• o ooo measured (Nimoh and Hanks, 1973b)
S
L,_ 4
o
n,- 3
2 o
o

% ,
,,,
>
0 i I
I00 200 300 400 500 600 o700 800 900

~-2 TIME (hrs)


o
~-3

Fig.4. Comparison b e t w e e n measured and predicted cumulative upward flow.


318

It should be noted that for the soil under consideration the calculated dif-
fusivity is not a monotonically increasing function of soil-water c o n t e n t
(Fig.5). For any grid node in which 0.33 < 0 < 0.365 cm 3 cm -3 , the value of
A0 and hence the selection of the appropriate Az becomes very important.
With a small change in Az, and hence AO, the value of OD(O)/az may vary
100
80
60

40

?
~ 2o

~ 6

,", 4

I i i i i J
0.1 0.2 0.3 0.4 0.5
SOIL-WATER CONTENT (cm3cm-3)
Fig.5. Relation between diffusivity a n d soil-water c o n t e n t for t h e soil u n d e r c o n s i d e r a t i o n .

from a very small negative number to zero or to a large positive number. This
in turn causes the calculated soil-water c o n t e n t to vary widely and the
numerical solution to become unreliable. Also, whenever, the nodal values
of soil-water content change rapidly due to a large surface flux, the selection
of the values of Az and A t becomes very important. In this study we have
used Az = 2.0 cm and At = 0.05 h. Although not included in Fig.5, it should
be noted that from step 2 of the computational procedure, diffusivities are cal.
culated for moisture contents ranging from 0.01 to 0.48 cm cm -3 . Values of
diffusivity for A0 = 0.01 covering the entire range of 0 are available from the
authors.
As part of the sensitivity of the model to various parameters, different
depths for the rooting zone were examined. The relative change of root den-
sity with depth was kept the same as t h a t for alfalfa given by Nimah and
Hanks (1973b). Fig.6 illustrates some numerical results for rooting depths
of 35, 60 and 70 cm. As the rooting zone decreases (keeping other variables
constant), soil-water content in the upper part of the soil profile also de-
creases. However, a decrease in the rooting zone causes almost no change in
deeper zones where the roots do not reach. In Fig.6a, for example, the upper
319

SOIL-WATER CONTENT {cm 3 cm-3)


O' O.I 0.2 0,3 0.4 0.5 0 0.1 O2 O3 O4 0.5
O. " ('/ b, ' ' '
2o \, '\, l /' 1-
(-f // /

E 60 \'\ i/
't/
x,,
I \<~I \
,,~iO0 '~~
140

Fig.& Soil-water content profiles predicted by the present model using a 70-era (solid
line), 50-era (broken line), and at-era (dots) rooting zone: (a) at the end of the irrigation
interval; and (b) following irrigation.

part of the soil with a 35-cm rooting zone is drier than that corresponding
to either a 50- or 70-cm rooting zone. Even though the drier soil initially
takes water at a greater rate, the wetting front advances at a lesser rate. Fig.
6b shows that for a rooting depth of 35 cm, where the soil is drier in the
upper part, the advance of the wetting front lags behind that for a rooting
depth of 50 cm. Similarly, the wetting front for a rooting depth of 50-cm
lags behind that for a rooting depth of 70 cm. This is consistent with the
results obtained by Warrick et al. (1971) in a flow model with no extraction
term.
To examine the response of the model to effective root density function,
two different types of functions given by Nimah and Hanks (1973b) and
suggested by Feddes et al. (1974) were used. Some of the calculated results
are illustrated in Fig.7 where the solid line represents the predicted soil-water
content profile using the t y p e of function suggested by Feddes et al. ([974)
and adjusted for alfalfa and the broken line represents the corresponding
profile using the r o o t density function given by Nimah and Hanks (1973b).
It is evident that the model is very sensitive to the type of root density func-
tion used. Thus, special attention must be given to the evaluation and use
of the appropriate function.
To examine the sensitivity ot the model to the depth of the lower boundary
(free water table), for predicting the amount of water flowing from or to the
lower boundary, depths of 82 and 164 cm have been considered (Fig.8).
After precipitation takes place (312--318 h and 613--627 h), the rate of
water movement to the lower boundary is higher for the 82-cm water table
than for the 164-cm water table. During periods of evaporation (no precipi-
tation), the movement of water from the water table to the r o o t zone is also
more pronounced for the case of the shallower water table.
320

SOIL-WATER CONTENT (cm 3 cm3)

2O
0 0.I
I '~''
I
03 0.5

"~ 6O

2
t---
w 100,
o

140

Fig.7. Soil-water content profile predicted by the present model using the type of root
d e n s i t y f u n c t i o n suggested b y F e d d e s et al. ( 1 9 7 4 ) a n d a d j u s t e d for alfalfa (solid line)
a n d N i m a h a n d H a n k s ( 1 9 7 3 b ) ( b r o k e n line) for a s i m u l a t e d flow t i m e o f 252 h.
6
5 - - 82 cm

12. 0 ~ L,. -- ----i

D I00 200 300 ~ " 600 / 700 800 900


LU - l i
_>
~-2
_J
TIME (hrs)
\
D- 3

e -4 i
-5
-6

Fig.8. Comparison of cumulative upward flow for two different depths to the lower
boundary.

SUMMARY AND CONCLUSIONS

A m a t h e m a t i c a l m o d e l has b e e n d e v e l o p e d t o p r e d i c t the soil-water c o n t e n t


profile u n d e r t r a n s i e n t field conditions. It takes into a c c o u n t a variable
t r a n s p i r a t i o n rate and a n o n u n i f o r m initial soil-water c o n t e n t . Rainfall, irri-
gation, and e v a p o r a t i o n are t r e a t e d as sources o f n o n u n i f o r m p o t e n t i a l sur-
face flux. T h e possibility o f n o flux at t h e soil surface is also i n c l u d e d in the
model. Plant c o n t r o l o n w a t e r u p t a k e w h e n soil w a t e r is a limiting f a c t o r is
n o t considered.
Solutions to the soft-water c o n t e n t m o d e l are o b t a i n e d n u m e r i c a l l y b y a
321

finite-difference scheme involving predictor--corrector equations. The scheme


results in a tridiagonal set of simultaneous equations which can be solved
rapidly in a digital computer. A step-by-step description of the computatio-
nal procedure is included.
For the soil under consideration, the calculated diffusivity is not a mono-
tonically increasing function of soft-water content. To obtain meaningful
soil-water content profiles, it is important to judiciously select the incremental
values of soil-water content and depth. Also, whenever the nodal values of
soil-water content change rapidly due to a large surface flux, the selection of
the incremental values of depth and time becomes very important. We have
used incremental values of 2.0 cm and 0.05 h for depth and time, respective-
ly. With these values simulation of 100 h required 3.5 min of processing time
on a Burroughs 6700 computer.
Application of the model to some field conditions and comparison of the
results with field-measured data showed very good agreement. The relative
deviation between measured and predicted values of the soil-water content
ranged from 0 to 21%.
When evaporation is considered as a source of flux at the soil surface, the
present model yields results that are in closer agreement to those measured
in the field than are those predicted by the models of Nimah and Hanks
(1973b) and Feddes et al. (1976). The sink term of Feddes et al. (1976) under-
estimates the soft-water content profile and in many cases does not yield re-
sults that are comparable to those measured in the field.
The agreement between measured and predicted soil-water content is
better when evaporation is included in the model. Thus, evaporation cannot
be disregarded in calculating soil-water content.
The model also has been used to calculate the amount of water flowing
from or to the b o t t o m boundary (free water table) during the period of flow.
The general trend for predicted flow through the b o t t o m boundary is the
same as that for measured flow. While the difference between predicted and
measured upward flow could be due to inaccuracy of the model and/or data,
the agreement is probably sufficiently good for many practical purposes.
Sensitivity of the model to changes in root zone, root density function,
and depth to the b o t t o m boundary has been analyzed. It has been shown that
as the rooting zone decreases (keeping other variables constant), soil-water
content in the upper part of the soil profile also decreases. However, a de-
crease in the rooting zone causes almost no change in deeper zones where the
roots do not reach. The response of the model to two different types of root
density functions indicates strong sensitivity to the type of root density
function used. Thus, special attention must be given to the evaluation and
use of the appropriate function. Finally, the rate of water movement from
or to a shallow water table is greater than that for a deep water table.
322

ACKNOWLEDGEMENT

The work reported herein was funded by CSRS Projects CA-D*-WSE-3081-H


and CA-D*-WSE-3479-H.

REFERENCES

Douglas, J. and Jones, B.F., 1963. On predictor--corrector methods for nonlinear parabolic
differential equations. J. Soc. Ind. Appl. Math., 11: 195--204.
Feddes, R.A., 1971. Water, heat and crop growth. Commun. Agric. Univ. Wageningen,
71.12, 184 pp.
Feddes, R.A., Bresler, E. and Neuman, S.P., 1974. Test of a modified numerical model
for water uptake by root systems. Water Resour. Res., 10(6): 1199--1206.
Feddes, R.A., Kowalik, P., Koliflska-Malinka, K. and Zaradny, H., 1976. Simulation of
field water uptake by plants using a soil water dependent root extraction function.
J. Hydrol., 31: 13--26.
Gardner, W.R., 1960. Dynamic aspects of water availability to plants. Soil Sci., 89(2):
63--73.
Gardner, W.R., 1964. Relation of root distribution to water uptake and availability. Agron.
J., 56: 41--45.
Gardner, W.R. and Ehlig, CoF., 1962. Some observations on the movement of water to
plant roots. Agron. J., 54: 453--456.
King, L.G. and Hanks, R.J., 1973. Irrigation management for control of quality of irrigation
return flow. Utah State Univ., Logan, Utah, Environ. Prot. Technol. Ser. EPA-R2-73-
265.
Molz, F.J. and Remson, I., 1970. Extraction term models of soil moisture use by trans-
piring plants. Water Resour. Res., 6(5): 1346--1356.
Molz, F.J., Remson, I., Fungaroli, A.A. and Drake, R.L., 1968. Soil moisture availability
for transpiration. Water Resour. Res., 4(6): 1161--1169.
Nimah, M.N. and Hanks, R.J., 1973a. Model for estimating soil water, plant, and atmo-
spheric interrelations, I. Description and sensitivity. Soil Sci. Soc. Am., Proc., 37 (4):
522--527.
Nimah, M.N. and Hanks, R.J., 1973b. Model for estimating soil water, plant, and atmo-
spheric interrelations, II. Field test of model. Soil Sci. Soc. Am., Proc., 37(4): 528--532.
Philip, J.R., 1957. The theory of infiltration. Soil Sci., 83: 345--357.
Warrick, A.W., Biggar, J.W. and Nielsen, D.R., 1971. Simultaneous solute and water trans-
fer for an unsaturated soil. Water Resour. Res., 7(5): 1216--1225.
Whisler, F.D., Klute, A° and Millington, R.M., 1968. Analysis of steady state evaporation
from soil column. Soil Sci. Soc. Am., Proc., 32(2): 167--174.

You might also like